Not content with space travel, electric cars and a social media platform, Elon Musk is now at the helm of a new online encyclopedia known as Grokipedia, which went live on Monday.
The Grok in the name refers to the generative AI chatbot from Musk’s startup xAI that features prominently on his social network, X. Grok will, according to the company, perform all fact-checking on Grokipedia.
But Grok, like other AI models of its type, has shared disinormation a number of times. In Grok’s case, this has included antisemitic tropes and praise for Adolf Hitler earlier this year.
Grokipedia v0.1 currently boasts just shy of 900,000 articles, around a tenth of Wikipedia. The crowd-edited, nonprofit online encyclopedia founded in 2001 has dominated the marketplace for more than two decades and is one of the most visited sites in the world.
Why has Elon Musk started Grokipedia?
Though he has previously praised Wikipedia, Musk has changed his position on the site of late. He labeled it “Wokepedia” in late September and accuses the site of favoring liberal causes and people in the way it presents information. Until recently, Musk was a friend of US President Donald Trump and publicly supported many of his conservative policies.
While there may be some commercial motivation for Grokipedia, Filippo Trevisan, an associate professor of public communication at American University in Washington, D.C., told DW the motivations are likely to be more ideological.
Grokipedia has about a tenth of the articles currently live on WikipediaImage: Hanno Bode/IMAGO
“Really money isn’t directly the objective here, this project really responds to those criticisms of Wikipedia from so many figures within the American conservative and the right-leaning world,” he said.
“There’s an attempt to try and provide an alternative to a traditional source, that might have been Wikipedia in this case, and try and make this outlet that’s AI-powered the new anchor, for alternative points of view.”
How is Grokipedia different to Wikipedia?
The distinctions are not always apparent to the user at first glance. In fact, Grokipedia leans heavily on Wikipedia for its information, as acknowledged at the bottom of some posts. Some entries are an exact copy of their Wikipedia counterparts, and the aesthetic is similar. Musk has said he wants this practice to end before 2026.
The main distinction between the two comes in how information is checked and processed. Wikipedia relies on collaborative community editing and has processes in place to flag any errors or points of contention. Roxana Radu, Associate Professor of Digital Technologies and Public Policy at the University of Oxford, UK, told DW that — with no human editorial involvement — Grokipedia appears to lack such oversight.
“Grokipedia operates on an obscure model of information gathering and sourcing, without transparency over the decisions taken ahead of displaying the content,” she said.
How accurate is Grokipedia?
Musk has claimed the new site “will exceed Wikipedia by several orders of magnitude in breadth, depth and accuracy” but given the high profile errors made by Grok and other AI models, there are serious doubts.
“Grokipedia deliberately pulls in sources that are not always pre-validated and might come accompanied by the label ‘fact‑checked by Grok,’ which has been contentious,” Radu said.
Radu said that her early impression of Grokipedia was “that it sometimes presents entries as a collage of discrete ideas, phenomena and concepts that are not always organized to provide a comprehensive overview on a topic, unlike Wikipedia.”
She added that “Grokipedia might leave its readers without the broader, holistic picture they would expect from a traditional encyclopaedia,” and gives more weight to things like Reddit posts and blog entries than traditional media.
AI is reshaping jobs, hiring, and productivity — here’s how
To view this video please enable JavaScript, and consider upgrading to a web browser that supports HTML5 video
While it is too early to assess Grokipedia’s accuracy on a macro scale, certain entries do omit important information. For instance, the article on Musk himself does not mention his hand gesture at a rally in January that many regarded as a Nazi salute, despite being significantly longer than its Wikipedia equivalent.
Is Grokipedia biased?
Musk’s belief, shared by many on the conservative side of US politics, is that traditional media and sources of official information are untrustworthy.
According to Trevisan, the success of the platform will depend on whether people buy the notion that AI is more impartial than humans.
“What Musk is trying to do is present AI as a solution to the bias problem,” Trevisan said. Musk appears to be “trying to capitalize on the impression that some might have, that taking the human element out might make this more objective.”
The lack of transparency in the process is also an issue for Trevisan: “We don’t know what goes on exactly in that black box, and so there isn’t the opportunity for us as consumers of information to verify why a piece of information might have ended up in the summary of an encyclopedia entry, which is the part that most people read most closely.”
Wikipedia itself has often faced accusations of bias, and not just from Musk and the American right. Studies by Havard University and the conservative-leaning Manhattan Institute think tank have found the site to show some evidence of left-wing bias: “Wikipedia entries are more likely to attach negative sentiment to terms representative of right-leaning political orientation than to their left-leaning counterparts,” concluded a report by Manhattan Institute expert David Rozado.
Both Radu and Trevisan agree that some bias is unavoidable in the provision of information, but that transparency, and admitting errors provide checks and balances which may not be employed by Grokipedia yet.
“What we can aspire to have is a balanced account, for which human interpretation is always needed,” concluded Radu.
Edited by: Carla Bleiker
